Efficient generation and execution of DAG-structured query graphs

Traditional database management systems use tree-structured query evaluation plans. While easy to implement, a tree-structured query evaluation plan is not expressive enough for some optimizations like factoring common algebraic subexpressions or magic sets. These require directed acyclic graphs (DAGs), i.e. shared subplans. This work covers the different aspects of DAG-structured query graphs. First, it introduces a novel framework to reason about sharing of subplans and thus DAG-structured query evaluation plans. Second, it describes the first plan generator capable of generating optimal DAG-structured query evaluation plans. Third, an efficient framework for reasoning about orderings and groupings used by the plan generator is presented. And fourth, a runtime system capable of executing DAG-structured query evaluation plans with minimal overhead is discussed. The experimental results show that with no or only a modest increase of plan generation time, a major reduction of query execution time can be achieved for common queries. This shows that DAG-structured query evaluation plans are serviceable and should be preferred over tree-structured query plans.